This is the fourth report in the 7-part series from The 20th Annual BCLT/BTLJ Symposium. The reports were written by Kevin Hickey, Microsoft Research Fellow, Berkeley Center for Law & Technology at Berkeley Law School.
Software Patents II: Williamson v. Citrix Online
A Report from the 20th Annual BCLT/BTLJ Symposium, Part 4
The Federal Circuit’s decision in Williamson v. Citrix Online, 792 F.3d 1339 (Fed. Cir. 2015) (en banc), represents a major development in the law of software patents. Understanding Williamson and predicting its effects was a thus central topic at the BCLT/BTLJ Symposium.
Williamson expanded the scope of “functional claiming” under 35 U.S.C. § 112(f). Section 112(f) allows a patentee to express her invention in functional terms—“a means for doing X”—instead of specifying a particular structure in the claims. The consequence, however, is that the patent claims are interpreted as covering only the structure in the patent specification that performs that function, and things like it—the “equivalents thereof.”
Before Williamson, whether a patent was subject to the costs and benefits of functional claiming was largely in the control of the patent drafter. Section 112(f) was presumed to apply only when the magic word “means” was used in the patent claims, and not otherwise. The Federal Circuit in Williamson eliminated this “strong” presumption against functional claiming, holding that claims will be interpreted as functional and subject to § 112(f) whenever the claim term fails to recite a sufficiently definite structure. In particular, the use of vague “nonce words” in patent claims—not just “means,” but terms like “module,” “mechanism”, “component”, et al.—is much more likely to trigger § 112(f) after Williamson.
Daniel Nazer of the Electronic Frontier Foundation reviewed the history that led to the Williamson decision. In Halliburton v. Walker, 329 U.S. 1 (1946), the Supreme Court invalidated a patent claim on § 112 grounds because one claim element was described only it terms of “what it will do,” and not in terms of its physical characteristics or structure. (Notably, this limitation was the point of novelty—the very thing that distinguished the Halliburton patent from the prior art.) Congress responded to Halliburton by enacting § 112(f), saving patents from invalidation if they used functional claiming, but limiting the scope of the patent claims in that case. This lead to two different theories of what Congress hoped to accomplish through § 112(f). Was it about patentee intent, that is, giving patentees the option to invoke functional claiming when they wished to avoid Halliburton? Or, instead, was it about restricting claim scope, such that claims must be tied to some actual structures disclosed in the patent (either in the claim itself or in the specification)? While pre-Williamson law favored the patentee intent theory, the Federal Circuit revered course with its en banc decision last year.
Sonal Mehta of Durie Tangri LLP looked at the effect of Williamson in district court decisions. Overall, Ms. Mehta saw a lot of variance in the results. In general, she found that the use of terms like “mechanism for,” “component for” and “device,” without more, triggered § 112(f). Courts have ruled both ways, however, with respect to “module.” (Of course, this approach of looking to the nonce words alone is obviously too simplistic—courts must to look at the whole claim in context. Still, it is an interesting empirical finding as a first-order approximation.) Usually, Williamson issues are fought at the claim construction stage, although sometimes defendants raise them via motions for summary judgment of patent indefiniteness.
Lee Van Pelt of Van Pelt, Yi & James LLP spoke to effects of Williamson on patent prosecution before the USPTO. For a long time, prosecutors tended to avoid writing § 112(f) claims, hoping to keep the claims as broad as possible. This strategy may end up backfiring, as Williamson gives patent defendants two potent arguments when § 112(f) is found to apply: (i) an indefiniteness argument if the specification fails to describe a structure corresponding to the function; and (ii) a noninfringement argument, because the claim scope is limited to equivalents of the disclosed structure. On the other hand, patentees may be able to use § 112(f) to their advantage to avoid § 101 problems, arguing that a broadly written claim is not a patent ineligible “abstract idea” under Alice, but really a narrower functional claim.
Going forward, Williamson has significantly changed the incentives for patent prosecutors, particularly for software patents. Mr. Van Pelt pointed to the USPTO training materials as a good resource for understanding how examiners will approach software claims subject to § 112(f) in patent prosecution. The essential procedure hinges on whether and how the patent discloses the algorithm at issue. If the algorithm is disclosed in the claim, §112(f) is unlikely to apply; if the algorithm is not disclosed in the claim, § 112(f) likely applies and the claim is limited to the algorithm in the specification and its equivalents; and if the software patent does not anywhere describe its algorithm, the patent is invalid. While these rules are reasonably straightforward in general, a big unresolved area is how they apply when § 112(f) is invoked in method claiming, i.e., where a patent claim describes a method step in functional terms.
Professor Kevin Collins argued that what he called the “Williamson revolution” will force the Federal Circuit to confront the fundamental question of what “structure” truly means in the context of software patents. In the mechanical arts, Williamson has had far less impact than in software, because everyone basically understands what “structure” means—physical objects. Software, however, is by its nature not tied to material objects, requiring a sui generis definition of “structure.” The Federal Circuit has proposed the term “algorithm” to fill this gap, but Professor Collins finds this definition unhelpful and even incoherent, since almost all claims recite a series of steps in some way. Professor Collins predicted that the tension Williamson created could lead to three possible resolutions: (1) all software claims will be subject to § 112(f), as by their nature they fail to recite any structure; (2) almost no software claims will be subject to § 112(f), because they literally recite a series of steps; or (3) the courts develop a better definition of what “structure” actually means in the software context.
Related Resources
Daniel Nazer – Williamson v. Citrix Online: The Demise, Return, and Second Demise(?) of Functional Claiming
Lee Van Pelt – Functional Claims
USPTO Examination Guidance and Training Materials – Evaluating § 112(f) Limitations in Software-Related Claims for Definiteness under 35 USC 112(b)
Kevin Emerson Collins, The Williamson Revolution in Software’s Structure, 30 Berkeley Tech. L.J. (forthcoming 2016) (draft)
Read more on the BCLT/BTLJ Software IP Symposium
This report is written by Kevin Hickey, the Microsoft Research Fellow at the Berkeley Center for Law & Technology, UC Berkeley School of Law. Mr. Hickey’s scholarship on intellectual property law and innovation policy is available online at SSRN.